Built by Exmergo · AI Agents for Your Data Stack.

PyPI License data-eng-bench ADE-bench CI

LinkedIn X

Install (Any Agent)

Run this command in your terminal

npx skills add exmergo/dex

Install (Claude Code)

Run these commands inside Claude Code one at a time

/plugin marketplace add exmergo/exmergo-agent-plugins

/plugin install dex@exmergo

Update later with /plugin marketplace update exmergo. The skills appear as /dex:explore, /dex:transform, and /dex:maintain and auto-trigger on matching intent. Ask it to warm dex once after installing, so the first real command does not wait for the engine to install (see Prerequisite: uv).

dex: the agent-native analytics engineering toolkit

dex is analytics engineering for Claude Code and any agent: data warehouse exploration, dbt transformation and semantic modeling, and schema-drift maintenance on dbt. Point it at your warehouse (or a local DuckDB file, or the one dex demo generates for you) and your dbt project; it learns the landscape, writes and refactors your dbt transformations and semantic models, and tells you what to fix when anything drifts. The dbt project is the source of truth; every change is a reviewable diff. Read-only against your data.

It closes the gap a general coding agent still has: agents re-learn the schema each session, have no strategy for thousands of tables, are blind to warehouse cost, will pull sensitive data into context, do not treat a dbt project as a first-class object, and have no concept of a semantic model to keep coherent over time. dex owns exactly that loop.

The loop

Explore. Transform. Maintain. (ETM)

  • Explore an unfamiliar warehouse: rank what matters, profile selectively, infer and verify joins, answer ad-hoc questions with guarded SQL probes behind a PII-aware query firewall, read the semantic layer as the object graph it is (semantic models, metrics with their composition, measures, dimensions, and the declared join graph, each resolved to the relation and column behind it, and the whole of it searchable and budgeted), read a dimension's value domain before filtering on it, and query its metrics (locally via MetricFlow or against a hosted dbt Cloud deployment), and render the map as a Mermaid ER diagram that draws the joins the semantic layer declares and never claims a cardinality the data has not proven. Persist a draft map. Fully read-only.
  • Transform the dbt project: author dbt models (staging to marts) with tests and docs, and the semantic layer on top (entities, dimensions, measures, metrics) as dbt semantic models (MetricFlow YAML), with a free Viz preview. Validated against a dev target, cost-guarded.
  • Maintain the project as it drifts: diff the warehouse and dbt against the last snapshot, surface schema, volume, grain, and definition drift ranked by blast radius, and propose edits.

Try it in three commands, on your laptop

No warehouse, no credentials, no cloud account, no network. dex demo generates a small e-commerce DuckDB warehouse locally and points the following commands at it.

pip install "exmergo-dex-core[duckdb]"
dex demo
dex explore map

dex demo writes two files in the directory you are standing in, and refuses rather than overwrite anything: dex_demo.duckdb (7 tables, 29,512 rows) and a .dex/config.yml so everything after it runs with no flags. The data is generated from a pinned seed, so what you see is what is written here.

It is seeded to be realistically broken, because a first run that reports a clean bill of health teaches you nothing. explore map flags 6 columns as personal data, infers 5 joins, and reports 5 data-quality findings. Then:

dex explore profile order_items products
dex explore relationships --verify
dex explore query "select email from customers"
  • A broken grain. order_item_id is not unique: 13000 distinct over 14000 rows, because a batch was loaded twice. Any join on it silently fans out.
  • A key that mixes id schemes. sku is 90% numeric, 10% 32-character hexadecimal (md5-shaped), from a merged catalogue. Cast it to a number and you drop 10% of your rows without an error.
  • A join that looks right and is not. web_events.customer_id shares the CRM's column name and type, so it is inferred; verification finds 100% of values have no match, so the inference collapses instead of shipping a join that returns all NULLs and looks like it worked.
  • A refusal. The query firewall declines to project customers.email into context. select count(distinct email) from customers runs, because a statistic is not a value.
  • Plus a table an interrupted load left empty, two columns whose declared type contradicts their content, and two PII false positives on a distribution centre's city and coordinates, which are a designed behaviour and worth meeting early.

DuckDB is free and local, so nothing here asks you to confirm a spend. On BigQuery or Snowflake the same commands return an estimate first and run only once you agree to it.

Prerequisite: uv

dex installs and runs its engine through uv, so you need it on your PATH before either install below. Neither Claude Code nor the plugin installs it for you.

curl -LsSf https://astral.sh/uv/install.sh | sh

brew install uv and pipx install uv work too. Nothing else is required: uv supplies the Python and the engine, with the connector extra chosen for you at runtime.

The first command in a fresh environment pays for that install, which is tens of seconds on a cold uv cache. --warm pays it up front instead: it materializes the environment, prints what it installed, and exits without running anything.

uv run --no-project --script <skill>/scripts/run.py --warm

Run it as a container build step or a CI setup step, or ask your agent to warm dex once after installing. Add --connector snowflake (or any other connector) to warm a warehouse before there is a project to read the choice from.

Benchmarks

We run dex on two public analytics-engineering benchmarks. Every run's raw per-task results are committed, including the ones that flatter us least.

data-eng-bench (Snowflake, 103 tasks)

Each task hands the agent a 2,356-model dbt project on a 489 MB DuckDB warehouse and a prescriptive ticket, then runs a hidden pytest suite after a cold dbt run. Scoring is binary and total: one failed assertion is a zero.

dex + Claude Sonnet 5 resolves 59 of 103 tasks (57.3%), with dex firing on 98% of trials. That is nominally the highest published Sonnet 5 figure and statistically indistinguishable from the 56.6% Snowflake published for both Claude Code and their own CoCo harness, since 0.7 points on 103 tasks is less than one task. Read it as parity, not as a win.

Because the reward is all-or-nothing, we also publish assertion-level results: 89.7% of assertions pass (task-weighted), and 19 of the 44 unresolved tasks missed by exactly one assertion. Full methodology and the per-check record are in the data-eng-bench README.

ADE-bench (dbt Labs, 75 tasks)

Fix, build, and extend dbt projects on DuckDB. dex + Claude Sonnet 5 reaches 76% task resolution, at 2.5x lower cost than Claude Fable 5.

With dex, accuracy clusters tightly across models (72-76%) while cost does not, so you can run an inexpensive model and still get top-tier results. One honest caveat on this one: dex was actually invoked in only 15 of the 75 Sonnet 5 trials, and on those 15 it netted a single extra task, so the 76% is mostly a statement about the model rather than about dex. Per-model cost and the raw results.json for every run are in the ADE-bench README.

On benchmarks

We publish these to be transparent, not to overclaim. A task-resolution score measures whether tests pass; it does not measure what matters most in practice: the experience of the human engineer working with the agent. Trust in a diff, clarity of the proposed change, cost surfaced before spend, and sensitive data kept out of context never show up in a pass rate. We optimize for that experience first and treat these scores as guide posts, not as the goal.

Two habits follow from that. We report how often dex actually ran next to the accuracy, because a score with the tool firing on 98% of trials and the same score with it firing on 20% are different claims and only the first says anything about dex. And we do not turn a gap smaller than the noise into a headline: a single run, one attempt per task, tells you roughly where a setup stands, not that it is better than the one a point below it.

Connectors

  • Cloud warehouse: Snowflake, BigQuery, Databricks, Amazon Redshift (Serverless-first), ClickHouse Cloud.
  • Self-hosted analytical: ClickHouse.
  • Embedded analytical: DuckDB.
  • Operational database: Postgres.

Credentials are discovered, never asked for: BigQuery through Application Default Credentials (gcloud auth application-default login), Snowflake through connections.toml, SNOWFLAKE_* env, or a dbt profile, Databricks through the SDK's unified chain (databricks auth login, DATABRICKS_* env, or a dbt profile), Redshift through the AWS credential chain (a pinned Serverless workgroup mints IAM temporary database credentials) or REDSHIFT_* env, Postgres through pg_service.conf, DATABASE_URL, the PG* environment, or a dbt profile, ClickHouse through CLICKHOUSE_URL, the CLICKHOUSE_* environment, or a dbt profile. Every scan is estimated and confirmed before it spends, capped server-side (maximum_bytes_billed on BigQuery; a per-statement statement timeout on Snowflake, Databricks, Redshift, and Postgres; max_execution_time plus max_bytes_to_read on ClickHouse). Budgets are bytes on BigQuery, warehouse-seconds with credits or DBUs alongside on Snowflake and Databricks, compute-seconds with RPU-hours alongside on Redshift, and database-seconds on Postgres and self-hosted ClickHouse. ClickHouse Cloud uses compute-seconds with live allocated memory translating to approximate compute-unit-hours and optional USD. All settled spend is recorded in a local ledger.

The two self-hosted connectors bill no dollars, and dex still gates them: an unbounded scan on a production Postgres primary or a shared ClickHouse cluster is a real cost even when nothing appears on an invoice.

Upcoming Connectors

  • Cloud warehouse: Trino, Azure Synapse, Microsoft Fabric

The exmergo-dex-core package

dex also bundles the exmergo-dex-core Python package.
This is the reusable and agent-friendly package that contains all the core explore, transform, and maintain logic. This also holds connectors and the write logic for .dex/ which stores cache, snapshots, and query billing logs.

You can install it yourself in your projects:

pip install exmergo-dex-core

or

uv add exmergo-dex-core

Use it as a library

The engine has a programmatic API, so you can drive the same loop from Python instead of shelling out to the CLI and parsing JSON:

from exmergo_dex_core import DexEngine

with DexEngine(connector="duckdb", path="shop.duckdb") as eng:
    mapped = eng.map()                  # a DexCache, not a JSON blob
    rows = eng.query("select status, count(*) from orders group by status")

print(mapped.relationship_count, rows.cells)

Methods return domain objects and result records, never the CLI's stdout envelope. Nothing above writes to disk: the default store keeps state in the process, so importing the package cannot leave a .dex/ directory in your repo. DexEngine.from_repo("path/to/repo") opts in to a project on disk, which is exactly what the CLI itself does.

Every guarantee the CLI makes holds here too, because it is the same code: reads are read-only, cost is surfaced before any spend (an unconfirmed billed call raises ConfirmationRequiredError carrying the estimate), and PII stays flagged rather than surfaced.

A process serving more than one end user can pass a ConnectionSource so each request reaches the warehouse as its own principal rather than as the container, and a SemanticSource to do the same for a hosted dbt Cloud Semantic Layer token. The host owns authentication; dex still builds the cost gate from your store, so the session budget binds either way.

More info in the package's README.md

Agent References

  • Cross-agent contract: AGENTS.md.
  • References (connectors, the contract, the canonical model, evaluation): references/.

Contributing

See CONTRIBUTING.md for local setup, the Ruff lint and format workflow, and the pre-commit hook. Every pull request into main must pass the Lint workflow and CI before it can merge.

Community

Community Integrations

dagster-dex

A rigorously-tested project format that allows dex to read a Dagster asset graph, maintained by CatInCloud Labs.
Instead of requiring a dbt project, dagster-dex maps your Dagster definitions directly into dex, allowing you to run schema-drift maintenance and exploration natively against your orchestrated assets.

Connect with the Community

Connect with the Analytics Engineering Community (Data Engineers welcome as well!) and discover how Exmergo brings AI Agents to Your Data Stack.

License

Apache-2.0.